Allow more PowerPC sibling calls

This patch enables sibling calls for powerpc in a few more cases, and fixes bugs exposed by that change. We now a) Allow sibling calls via function pointer. At the time rs6000_function_ok_for_sibcall was written, I don't think access to arg types of function pointer calls was available in t

[pph] Debug Cleanup (issue4380046)

Move use of flag_pph_debug to the new flag_pph_decls_debug for the declaration dependence analysis. This code should be dead, but we'll preserve until we know for sure. Split pph_output_tree into pph_output_tree and pph_output_tree_aux. The former is the primary external interface, and pph_outpu

fix testsuite/48506

I've checked in this to avoid problems with ports that default to -fno-common. I've also checked it into 4.6.x. PR testsuite/48506 * gcc.dg/tree-ssa/ssa-ccp-17.c: Robustify against ports that default to fno-common. Index: gcc.dg/tree-ssa/ssa-ccp-17.c

Re: [testsuite] fix to gcc.target/arm/pr43698.c

On Apr 8, 2011, at 8:07 AM, Janis Johnson wrote: > Test gcc.target/arm/pr43698.c specifies -march=armv7-a and fails > execution for multilibs whose hardware or simulator doesn't support that > architecture. Ideally, I'd like target people to weigh in on target changes, ssa people to weigh in on s

RE: [Ping][Patch, testsuite]: Don't xfail sibcalls on AVR

> -Original Message- > From: Georg-Johann Lay [mailto:a...@gjlay.de] > Sent: Tuesday, March 29, 2011 3:00 AM > To: gcc-patches@gcc.gnu.org > Cc: Anatoly Sokolov; Denis Chertykov; Weddington, Eric > Subject: [Ping][Patch, testsuite]: Don't xfail sibcalls on AVR > > Target avr now supports

Small tweak to combine_validate_cost

At the end of combine_validate_cost, when the combination is validated, the registered cost of I1 is zero-ed. This should be done for I0 as well. Bootstrapped/regtested on x86_64-suse-linux, applied on the mainline. 2011-04-08 Eric Botcazou * combine.c (combine_validate_cost): Adju

Fix PR48484

Hi, this patch fixes the test failure. OK after bootstrap and regression test? Thanks, David 2011-04-08 Xinliang David Li PR tree-optimization/PR48484 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Set has_valid_pred lazily 2011-04-08 Xinliang David Li

Re: [patch, fortran] More control over front end optimization

Hello Mikael and Daniel, thanks for the reviews. Here is what I committed as rev. 172215. I tried to be a little bit more specific in the doc, but couldn't think of a better name for the options without trying to win a prize for Gcc's longest option name. Thomas 2011-04-08 Thomas Koen

Re: FDO usability -- sanity check indirect call target

Please review this patch. Regression test is ok. I will do more application testing to make sure the check is not too strict (filtering out legal ic targets). Thanks, David 2011-04-07 Xinliang David Li * value-profile.c (function_decl_num_args): New function. (check_ic_target

Re: [PATCH] Fix LTO bootstrap on i686-linux (problem with two Ldebug_info0 labels; PR bootstrap/48148)

Hi, On Fri, 8 Apr 2011, Richard Guenther wrote: > > > Sounds like this comment needs to be updated if there can be types > > > on the list as well. > > > > On a closer look, this seems to be because LTO messes up types > > terribly, struct cpp_options's lang field doesn't have enum c_lang > >

Implement stack arrays even for unknown sizes

Hello, I developed this patch during stage3 of 4.6, so it wasn't appropriate then, but now should be a good time. It adds a new option -fstack-arrays which makes the frontend put all local arrays on stack memory. Even those of non-constant size, by using alloca. Or rather not by explicitely

Re: [PATCH] Fix LTO bootstrap on i686-linux (problem with two Ldebug_info0 labels; PR bootstrap/48148)

On Fri, 8 Apr 2011, Jakub Jelinek wrote: > On Fri, Apr 08, 2011 at 01:58:14PM -0400, Jason Merrill wrote: > > On 04/05/2011 10:19 AM, Jakub Jelinek wrote: > > >i686-linux LTO bootstrap currently fails, because in one partition > > >we emit .Ldebug_info0 label twice. The problem is that > > >resol

Re: [patch i386]: Cleanup calling convention handling in i386.c and fix PR target/9601 and PR target/11772

> + /* Special case regparm/sseregparm, which are either cdecl or stdcall. */ > + if ((ret & (IX86_CALLCVT_REGPARM | IX86_CALLCVT_SSEREGPARM)) != 0) > +return (ret | ((TARGET_RTD && !stdarg_p (type)) ? IX86_CALLCVT_STDCALL > + : IX86_CALLCVT_

[Ada] improve layout of RETURN type made for CICO mechanism

One year ago the compiler (gigi) was changed to align the RETURN structure generated for the Copy-In Copy-Out mechanism: http://gcc.gnu.org/ml/gcc-patches/2010-04/msg00256.html This helps to generate better code, especially on 64-bit platforms. But this is not sufficient in some cases, i.e. whe

[Ada] Fix ICE on array type with aggregate component as bound

The compiler aborts on the declaration of an array type whose upper bound is a component of an aggregate whose nominal subtype is a discriminated record type which contains a component whose nominal subtype is a volatile discriminated record type with a variant part and a size clause. Tested on

Re: [pph] Macro Validation (issue4379044)

> "Lawrence" == Lawrence Crowl writes: Diego> Hm, and 5 bits are still enough for directive_index? Lawrence> Yes there are enough. I added a comment at the table definition Lawrence> to ensure that future changes keep it so. I suggest a static assertion of some kind. Or in lieu of that, m

Re: Ping Re: Target header etc. cleanup patch

On Apr 8, 2011, at 11:46 AM, Joseph S. Myers wrote: > Ping. This patch > is pending > 011-04-02 Joseph Myers This is wrong, it should use @ instead of :. :-) I glanced at the rest (in part to see if there were any darwin bits for

Re: [PATCH] Fix LTO bootstrap on i686-linux (problem with two Ldebug_info0 labels; PR bootstrap/48148)

On Fri, Apr 08, 2011 at 01:58:14PM -0400, Jason Merrill wrote: > On 04/05/2011 10:19 AM, Jakub Jelinek wrote: > >i686-linux LTO bootstrap currently fails, because in one partition > >we emit .Ldebug_info0 label twice. The problem is that > >resolve_addr for call_site support attempts to force_decl

Ping Re: Target header etc. cleanup patch

Ping. This patch is pending review. The updated version (with one bit that had been committed separately removed) applies cleanly to current trunk. -- Joseph S. Myers jos...@

Re: [pph] Macro Validation (issue4379044)

On 4/8/11, Diego Novillo wrote: > On Thu, Apr 7, 2011 at 13:54, Lawrence Crowl wrote: >> *** pth_save_token_cache (cp_token_cache *ca >> *** 448,453 >> --- 448,454 >>pph_output_uint (f, num); >>for (i = 0, tok = cache->first; i < num; tok++, i++) >> pth_save_tok

Re: [PATCH] Fix LTO bootstrap on i686-linux (problem with two Ldebug_info0 labels; PR bootstrap/48148)

On 04/05/2011 10:19 AM, Jakub Jelinek wrote: i686-linux LTO bootstrap currently fails, because in one partition we emit .Ldebug_info0 label twice. The problem is that resolve_addr for call_site support attempts to force_decl_die external function decls, and at least with LTO that in turn can att

Re: PATCH [trunk] gengtype should generate ggc_alloc macros in plugin mode.

On Fri, 8 Apr 2011 15:29:48 +0300 Laurynas Biveinis wrote: > > 2011-04-08  Basile Starynkevitch   > >        * gengtype.c (write_typed_alloc_def): New argument f. Use it instead > >          of header_file. > >          (write_typed_struct_alloc_def, write_typed_typedef_alloc_def) > >          (w

Re: [PATCH PING] c++-specific bits of tree-slimming patches

On 03/24/2011 09:15 AM, Nathan Froyd wrote: The C++-specific bits of these patches: [PATCH 02/18] enforce TREE_CHAIN and TREE_TYPE accesses http://gcc.gnu.org/ml/gcc-patches/2011-03/msg00557.html OK. [PATCH 07/18] generalize build_case_label to the rest of the compiler http://gcc

Re: patch to fix PR 48435

On 04/08/2011 01:28 PM, Graham Stott wrote: Vladimir, The wrong PR 4435 was in referrenced the commit Thanks, Graham. I've just fixed it. I should change my keyboard. This is not the first time when some symbols are missed.

Re: patch to fix PR 48435

Vladimir, The wrong PR 4435 was in referrenced the commit --- On Fri, 8/4/11, Jeff Law wrote: > From: Jeff Law > Subject: Re: patch to fix PR 48435 > To: "Vladimir Makarov" > Cc: "gcc-patches" > Date: Friday, 8 April, 2011, 16:18 > -BEGIN PGP SIGNED MESSAGE- > Hash: SHA1 > > On 04/

[committed] Fix secondary reload support for PA shift amount register

Recent changes to the ira code exposed problems in handling secondary reloads for copies between the shift amount register and floating point registers. This requires an intermediate general register. I also saw a case where reload tried to stuff a SFmode value into the shift amount register comp

Re: Simple enhancement to -dA dump

On Fri, Apr 8, 2011 at 12:20, Xinliang David Li wrote: >>>   /* Output the insns.  */ >>>   for (insn = first; insn;) >>>     { >>> @@ -1721,8 +1788,19 @@ final (rtx first, FILE *file, int optimi >>>        insn_current_address = INSN_ADDRESSES (INSN_UID (insn)); >>>  #endif /* HAVE_ATTR_length *

Re: Simple enhancement to -dA dump

On Fri, Apr 8, 2011 at 5:04 AM, Diego Novillo wrote: > On 04/03/2011 08:54 PM, Xinliang David Li wrote: > >> + >> +      fprintf (file, "# BLOCK %d", bb->index); >> +      if (bb->frequency) >> +        fprintf (file, " freq:%d", bb->frequency); >> +      if (bb->count) >> +        fprintf (file,

[PATCH][RFC] Make called function type explicit, make function pointer type conversions useless

This is the "real" fix for PR46076 that I wanted to persue. Make function pointer type conversions useless as to more aggressively be able to turn indirect into direct calls. This requires that we preserve the original type signature of the called function as presented by the frontend. The patc

Re: [C++0x] Range-based for statements and ADL

Looks good, just a couple of tweaks: On 04/06/2011 07:22 PM, Rodrigo Rivas wrote: + error ("range-based % expression must have complete type"); + error ("range-based % expression has an % member " + "but not a %"); Let's give the type of the range initializer

Re: [PATCH] use use_reg in several backends

On 04/05/2011 10:55 AM, Nathan Froyd wrote: > * config/alpha/alpha.c (alpha_emit_xfloating_libcall): Call use_reg. > * config/xtensa/xtensa.c (xtensa_call_tls_desc): Likewise. > * config/sparc/sparc.c (sparc_legitimize_tls_address): Likewise. Ok. r~

Re: patch to fix PR 48435

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 04/07/11 13:41, Vladimir Makarov wrote: > The following patch should solve problem > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48435. It might solve other > latest IRA problems too including performance related ones. The patch > is for targets w

[testsuite] fix to gcc.target/arm/sync-1.c

Test gcc.target/arm/sync-1.c specifies -march=armv7-a and fails execution for multilibs whose hardware or simulator doesn't support that architecture. The test uses __sync_fetch_and_add, which GCC doesn't support for all arm targets. Using the -march option allows it to compile if the multilib op

[testsuite] fix to gcc.target/arm/pr43698.c

Test gcc.target/arm/pr43698.c specifies -march=armv7-a and fails execution for multilibs whose hardware or simulator doesn't support that architecture. The test doesn't depend on that option, and without it will occasionally run for such a target anyway. This patch removes the option, letting it

Re: Fix PR48389: ICE in make_edges

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 04/08/11 07:33, Michael Matz wrote: > Hi, > > the problem is that some initializers expand to loops. If such > initializers are inserted on edges we have looping control flow on them. > Even if we have no loops, but normal control flow on them

RE: CR16 Port addition

This is not a patch against GCC trunk. Please try again with a patch that is diffed against current GCC SVN trunk and that builds cleanly (including building with --enable-werror-always, starting from a native bootstrapped build of current trunk as the compiler you use to compile your cross co

Re: Fix PR48389: ICE in make_edges

On Fri, Apr 8, 2011 at 6:58 AM, Michael Matz wrote: > Hi, > > On Fri, 8 Apr 2011, Jakub Jelinek wrote: > >> On Fri, Apr 08, 2011 at 03:33:49PM +0200, Michael Matz wrote: >> > --- testsuite/gcc.target/i386/pr48389.c     (revision 0) >> > +++ testsuite/gcc.target/i386/pr48389.c     (revision 0) >> >

Re: [gc-improv] Permanent vs function RTL obstack fix

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 04/08/11 08:39, Richard Guenther wrote: > On Fri, Apr 8, 2011 at 4:36 PM, Jeff Law wrote: >> -BEGIN PGP SIGNED MESSAGE- >> Hash: SHA1 >> >> On 04/08/11 07:21, Laurynas Biveinis wrote: >>> 2011/4/8 Steven Bosscher : >* stmt.c (la

Re: [gc-improv] Permanent vs function RTL obstack fix

On Fri, Apr 8, 2011 at 4:36 PM, Jeff Law wrote: > -BEGIN PGP SIGNED MESSAGE- > Hash: SHA1 > > On 04/08/11 07:21, Laurynas Biveinis wrote: >> 2011/4/8 Steven Bosscher :        * stmt.c (label_rtx): Allocate RTX in permanent RTL memory. >>> >>> That looks strange, labels should be funct

Re: [gc-improv] Permanent vs function RTL obstack fix

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 04/08/11 07:21, Laurynas Biveinis wrote: > 2011/4/8 Steven Bosscher : >>>* stmt.c (label_rtx): Allocate RTX in permanent RTL memory. >> >> That looks strange, labels should be function specific, except >> non-local labels. Maybe you can get

[PATCH] Remove kludge in gimple_call_flags

Now as we don't have to handle removed stmts all the time remove this kludge. Bootstrapped and tested on x86_64-unknown-linux-gnu, applied. Richard. 2011-04-08 Richard Guenther * gimple.c (gimple_call_flags): Remove kludge. Index: gcc/gimple.c ==

[committed] sel-sched: Fixup adding of preheader blocks from inner loops

I failed to notice that current_loop_nest is modified in setup_current_loop_nest and used in sel_setup_region_sched_flags. Therefore, sel_setup_region_sched_flags should be moved with setup_current_loop_nest after sel_init_bbs as well. I've committed the following patch under the 'obvious' rule

Re: [patch libjava]: Do some fixes for windows native targets

2011/4/8 Tom Tromey : >> "Kai" == Kai Tietz writes: > > Kai> I added to headers.txt the changes and regenerate headers.  Btw it is > Kai> a bit tricky to enable for libjava the maintainer-mode and then have a > Kai> successful configure for it ... > > [...] > > Kai> Sorry, found a typo. Update

Re: Fix PR48389: ICE in make_edges

On Fri, Apr 08, 2011 at 03:58:57PM +0200, Michael Matz wrote: > But then I'd have to use --target_board to hit the original problem. That isn't so a big deal, testing with RUNTESTFLAGS='--target_board=unix\{-m32,-m64\}' is what people do very often. Or just do a 32-bit instead of 64-bit bootstr

Re: Fix PR48389: ICE in make_edges

Hi, On Fri, 8 Apr 2011, Jakub Jelinek wrote: > On Fri, Apr 08, 2011 at 03:33:49PM +0200, Michael Matz wrote: > > --- testsuite/gcc.target/i386/pr48389.c (revision 0) > > +++ testsuite/gcc.target/i386/pr48389.c (revision 0) > > @@ -0,0 +1,12 @@ > > +/* PR middle-end/48389 */ > > +/* { dg-d

Re: [patch libjava]: Do some fixes for windows native targets

> "Kai" == Kai Tietz writes: Kai> I added to headers.txt the changes and regenerate headers.  Btw it is Kai> a bit tricky to enable for libjava the maintainer-mode and then have a Kai> successful configure for it ... [...] Kai> Sorry, found a typo. Updated patch attached. This is ok. Than

Re: Fix PR48389: ICE in make_edges

On Fri, Apr 08, 2011 at 03:33:49PM +0200, Michael Matz wrote: > --- testsuite/gcc.target/i386/pr48389.c (revision 0) > +++ testsuite/gcc.target/i386/pr48389.c (revision 0) > @@ -0,0 +1,12 @@ > +/* PR middle-end/48389 */ > +/* { dg-do compile } */ > +/* { dg-options "-O -m32 -mtune=penti

Re: [pph] Macro Validation (issue4379044)

On Thu, Apr 7, 2011 at 13:54, Lawrence Crowl wrote: > *** pth_save_token_cache (cp_token_cache *ca > *** 448,453 > --- 448,454 >    pph_output_uint (f, num); >    for (i = 0, tok = cache->first; i < num; tok++, i++) >      pth_save_token (tok, f); > + >  } Extra blank line.

Fix PR48389: ICE in make_edges

Hi, the problem is that some initializers expand to loops. If such initializers are inserted on edges we have looping control flow on them. Even if we have no loops, but normal control flow on them (which can happen in other situations) we still forget to initialize JUMP_LABEL on them leadin

[PATCH, committed] Fix PR 48272

Hello, This is the case when sched-pressure support is not ready for new insns appearing during scheduling. (Found with the crazy set of options on ia64, where register pressure is not much of an issue.) Anyways, the fix is simply initializing the proper data structures at the point where ne

Re: [gc-improv] Permanent vs function RTL obstack fix

2011/4/8 Steven Bosscher : >>        * stmt.c (label_rtx): Allocate RTX in permanent RTL memory. > > That looks strange, labels should be function specific, except > non-local labels. Maybe you can get away with allocating DECL_NONLOCAL > labels on the permanent rtl obstack? That's a good idea, I

New German PO file for 'gcc' (version 4.6.0)

Hello, gentle maintainer. This is a message from the Translation Project robot. A revised PO file for textual domain 'gcc' has been submitted by the German team of translators. The file is available at: http://translationproject.org/latest/gcc/de.po (This file, 'gcc-4.6.0.de.po', has just

[PATCH] Move MODIFIED_NORETURN_CALLS queueing

This moves queueing calls to MODIFIED_NORETURN_CALLS from gimple_set_modified to update_stmt_operands. The former is called when we remove stmts and thus the call may reference freed SSA names which doesn't make gimple_call_noreturn_p happy (and in fact we have a kludge to avoid ICEing here). A

Re: [PATCH] Fix PR48467

On Wed, Apr 6, 2011 at 11:47, Richard Guenther wrote: > 2011-04-06  Richard Guenther   > >        PR lto/48467 >        * toplev.c (lang_dependent_init): Do not open asm_out_file >        in WPA mode, nor perform debug machinery initialization. >        (finalize): Do not unlink asm_out_file in W

Re: PATCH [trunk] gengtype should generate ggc_alloc macros in plugin mode.

> 2011-04-08  Basile Starynkevitch   >        * gengtype.c (write_typed_alloc_def): New argument f. Use it instead >          of header_file. >          (write_typed_struct_alloc_def, write_typed_typedef_alloc_def) >          (write_typed_alloc_defns): Likewise. >          (main): Calls write_typed

[PATCH] Abstract call stmt function type

This prepares a patch that makes the type of the called function explicit in gimple calls rather than taking this information from the called function pointer. It introduces a gimple_call_fntype accessor that the replaces TREE_TYPE (TREE_TYPE (gimple_call_fn ( pattern used everywhere (which a

Re: Simple enhancement to -dA dump

On 04/03/2011 08:54 PM, Xinliang David Li wrote: + + fprintf (file, "# BLOCK %d", bb->index); + if (bb->frequency) +fprintf (file, " freq:%d", bb->frequency); + if (bb->count) +fprintf (file, " count:" HOST_WIDEST_INT_PRINT_DEC, + bb->count); +

[committed] Fix vectorization with non-naturally aligned accesses (PR tree-optimization/48377)

Hi! Packed attribute/pragma isn't the only way to cause types to be less than naturally aligned. Aligned attribute can do the same thing (and, for non-aggregates, packed attribute is ignored and thus aligned(X) is the only way to express it). The alignment_reachable targhook takes (or should take

[PATCH] Fix PR48513

I've re-generated tm.texi and verified make doc now works. Installed. Richard. 2011-04-08 Richard Guenther PR bootstrap/48513 * doc/tm.texi: Re-generate. Index: gcc/doc/tm.texi === --- gcc/doc/tm.texi (revi

Re: [PATCH, ARM] PR47855 Compute attr "length" for some thumb2 insns, 2/3

On 08/04/11 10:57, Carrot Wei wrote: Hi This is the second part of the fixing for http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47855 This patch contains the length computation for insn patterns "*arm_movqi_insn" and "*arm_addsi3". Since the alternatives and encodings are much more complex, the

[PATCH, ARM] PR47855 Compute attr "length" for some thumb2 insns, 2/3

Hi This is the second part of the fixing for http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47855 This patch contains the length computation for insn patterns "*arm_movqi_insn" and "*arm_addsi3". Since the alternatives and encodings are much more complex, the attribute length is computed in separat

Re: [Ping #2][Patch, testsuite]: Don't xfail sibcalls on AVR

Gerald Pfeifer schrieb: > On Wed, 6 Apr 2011, Georg-Johann Lay wrote: >> I signed a fsf copyright assignment. AFAIK there is no automatism >> that grants write access when copyright assignment is approved. > > Yes, this is correct. > >> I would have to be maintainer or something like that. > > N

Re: FDO usability -- sanity check indirect call target

On Fri, Apr 8, 2011 at 8:06 AM, Xinliang David Li wrote: > Hi, due to race conditions, it is common that the value profile > information for an indirect call site is 'corrupted' -- resulting in > false target function to be recorded. The value profile transformation > won't cause runtime problem a

Re: FDO usability patch -- correct insane profile data

On Fri, Apr 8, 2011 at 1:39 AM, Xinliang David Li wrote: > Hi please review the attached patch. > > Ok when bootstrap and test finish? Ok. Thanks, Richard. > Thanks, > > David > > > > 2011-04-07  Xinliang David Li   > >        * ipa-cp.c (ipcp_update_profiling): Correct >        negative scale

RFC: Gimple calls to "internal" functions

I'm implementing the multi-vector load/store support described here: http://gcc.gnu.org/ml/gcc/2011-03/msg00322.html For the purposes of this message, the main thing is that the losds and stores are more naturally represented as GIMPLE_CALLs rather than GIMPLE_ASSIGNs. That is, we have:

Re: [PATCH, ARM] PR47855 Compute attr "length" for some thumb2 insns

On 8 April 2011 09:52, Carrot Wei wrote: > Sorry, which pattern in thumb2.md? Sorry, I meant to say push_multi in arm.md . Not enough coffee yet this morning. cheers Ramana > > thanks > Carrot > > On Fri, Apr 8, 2011 at 4:35 PM, Ramana Radhakrishnan > wrote: >> On 08/04/11 08:36, Carrot Wei wro

Re: [PATCH, ARM] PR47855 Compute attr "length" for some thumb2 insns

Sorry, which pattern in thumb2.md? thanks Carrot On Fri, Apr 8, 2011 at 4:35 PM, Ramana Radhakrishnan wrote: > On 08/04/11 08:36, Carrot Wei wrote: >> >> Hi >> >> This patch moves the length computation of push_multi into a C >> function to avoid the usage of GNU extension. > > Please put the co

Re: [PATCH, ARM] PR47855 Compute attr "length" for some thumb2 insns

On 08/04/11 08:36, Carrot Wei wrote: Hi This patch moves the length computation of push_multi into a C function to avoid the usage of GNU extension. Please put the comment regarding the calculation of the length attribute along with the pattern in thumb2.md as well. Tested on qemu without

[patch, ARM] PR48250, rehaul arm_legitimize_reload_address()

Hi Richard, here's a new patch, with some quite significant changes to arm_legitimize_reload_address(). This time fully utilizing the sign-magnitude offset macro you gave in the last thread, and adapting your description into embedded comments. The reload_outdf pattern caused some regressions in a

Re: [PATCH, ARM] PR47855 Compute attr "length" for some thumb2 insns

Hi This patch moves the length computation of push_multi into a C function to avoid the usage of GNU extension. Tested on qemu without regression. OK to install? thanks Carrot ChangeLog: 2011-04-08 Wei Guozhi PR target/47855 * config/arm/arm-protos.h (arm_attr_length_push_mu